A segmentation method for nuclei identification from sagittal images of Drosophila melanogaster embryos

Abstract

This paper proposes a segmentation method for sagittal images obtained from Drosophila melanogaster embryos at early stages of development. The proposed method operates in the spatial domain and uses traditional filters and segmentation techniques for image processing. However, one common problem encountered after the segmentation of sagittal images is the merged nuclei. In order to split these nuclei, our approach uses the curvature of the shape of the merged nuclei to find the regions of division between the nuclei. The proposed method was compared to other techniques and it achieved better results. In general, the obtained results have shown good performance and were able to identify individual nuclei, providing an efficient and accurate solution for segmentation of nuclei in images obtained from the sagittal plane of Drosophila embryos.